SILC · Silicom Ltd.

Silicom Ltd. announced that a pioneering AI inference acceleration provider has selected its inference-specific solution for a Proof of Concept (PoC) targeting mainly Tier-1 hyperscalers. The customer placed initial orders for delivery in the first half of 2026, with the PoC scheduled for the second half of 2026. Upon successful completion, the customer anticipates transitioning to an initial full-scale deployment requiring tens of thousands of Silicom units, each with a multi-thousand-dollar Average Selling Price (ASP). This milestone underscores Silicom's rapid progress in the AI inference market, leveraging adaptable FPGA and COTS networking chip architectures to overcome latency and hardware lottery risks. The company has now secured orders from two of the world's most promising AI compute contenders for two distinct inference products, along with a recent order for a third inference-specific product. CEO Liron Eizenman highlighted a tectonic shift from AI training to massive inference deployments, where purpose-built architectures offer lower total cost of ownership per token and superior energy efficiency. These early strategic engagements validate Silicom's technological edge and could significantly enhance its long-term growth trajectory if successfully capitalized upon.